A dynamic random access memory (DRAM) cell typically comprises a charge storage capacitor (or cell capacitor) coupled to an access device such as a Metal-Oxide-Semiconductor Field Effect Transistor (MOSFET). The MOSFET functions to apply or remove charge on the capacitor, thus affecting a logical state defined by the stored charge. The amount of charge stored on the capacitor is determined by the capacitance, C=.epsilon.A/d, where .epsilon. is the dielectric constant of the capacitor dielectric, A is the electrode (or storage node) area and d is the interelectrode spacing. The conditions of DRAM operation such as operating voltage, leakage rate and refresh rate, will in general mandate that a certain minimum charge be stored by the capacitor.
In the continuing trend to higher memory capacity, the packing density of storage cells must increase, yet each will maintain required capacitance levels. This is a crucial demand of DRAM fabrication technologies if future generations of expanded memory array devices are to be successfully manufactured. Nevertheless, in the trend to higher memory capacity, the packing density of cell capacitors has increased at the expense of available cell area. For example, the area allowed for a single cell in a 64-Mbit DRAM is only about 1.4 .mu.m.sup.2. In such limited areas, it is difficult to provide sufficient capacitance using conventional stacked capacitor structures. Yet, design and operational parameters determine the minimum charge required for reliable operation of the memory cell despite decreasing cell area. Several techniques have been developed to increase the total charge capacity of the cell capacitor without significantly affecting the cell area. These include new structures utilizing trench and stacked capacitors, electrodes having textured surface morphology and new capacitor dielectric materials having higher dielectric constants.
Recently, for example, a great deal of attention has been given to the development of thin film dielectric materials that possess a dielectric constant significantly greater (&gt;10.times.) than the conventional dielectrics used today, such as silicon oxides or nitrides. Particular attention has been paid to Barium Strontium Titanate (BST), Barium Titanate (BT), Lead Zirconate Titanate (PZT), Tantalum Pentoxide (Ta.sub.2 O.sub.5) and other high dielectric constant materials as a cell dielectric material of choice for DRAMs. These materials, in particular BST, have a high dielectric constant (&gt;300) and low leakage currents which makes them very attractive for high density memory devices. Due to their reactivity and complex processing, these dielectric materials are generally not compatible with the usual polysilicon electrodes. Thus, much effort has been directed to developing suitable metal electrodes for use with such dielectric materials.
As DRAM density has increased (1 MEG and beyond), thin film capacitors, such as stacked capacitors (STC), trenched capacitors, or combinations thereof, have evolved in attempts to meet minimum space requirements. Many of these designs have become elaborate and difficult to fabricate consistently as well as efficiently. Furthermore, the recent generations of DRAMs (4 MEG, 16 MEG for example) have pushed conventional thin film capacitor technology to the limit of processing capability. In giga-scale STC DRAMs the electrode conductivity plays an important role in device size and performance; thus, two kinds of capacitors have been considered, the three-dimensional metal electrode such as the FIN or CROWN, or the simple metal electrode with higher-permitivity dielectric films. For example, a recent article by T. Kaga et al. ("0.29 .mu.m.sup.2 MIM-CROWN Cell and Process Technologies for 1-Gigabit DRAMs," T. Kaga et al., IEDM '94, pp. 927-929.) discloses a substituted tungsten process for forming three-dimensional metal electrodes from polysilicon "molds." The article, herein incorporated by reference, discloses a method advantageous for creating metal structures, such as capacitor electrodes; nevertheless the simple structures created thus far are not sufficient to meet the demands of giga-scale DRAM arrays.